With traditional T1- and T2-weighted spin echo (SE) sequences, moving blood appears homogeneously hypointense or black as the excited spins exit the imaged slice. This exit phenomenon yields no signal and is commonly termed as flow void (21). With flow void, intraluminal thrombus is appreciated as signal residing within a vessel lumen. Thrombus signal intensity depends on age of the thrombus but does not necessarily follow the same predictable time-dependent appearance as intracranial hemorrhage. Slow flow in a patent vessel can also demonstrate increased signal (21). Conventional T1- and T2-weighted SE sequences are not recommended as a first-line strategy for chest and abdominal MRV, as these require long acquisition times.

Fast spin echo (FSE) sequences (or turbo SE sequences) can be used to decrease long acquisition times associated with conventional SE sequences. For FSE sequences, multiple 180-degree refocusing pulses are applied after each 90-degree pulse, and the subsequent echoes are used to fill in the lines of k-space sequentially (22). With multiple phase-encoded steps performed per repetition time (TR), the speed of sequence acquisition is reduced proportional to the echo train length.

It is possible to use FSE to exploit the long T2 characteristics of blood and provide bright signal within blood vessels, independent from the direction or velocity of the flow (23). The T2 of blood depends on its oxygenation level and thus varies from 60 milliseconds in venous blood (oxygenation between 60% and 75%) to 220 milliseconds for arterial blood (24).

The T2 of venous blood is substantially longer than that of the other tissues on FSE acquisitions with long tau (time between 180° refocusing pulses) (25). A short echo time (TE) (<100 milliseconds) and a longer inter-TE can augment the signal difference between venous and arterial blood (25). At 1.5 T, a TE of 50 milliseconds provided maximum contrast-to-noise in the calf veins using a fat-saturated FSE sequence (23) although a TE >60 milliseconds can offer adequate suppression of the muscle (25). Fat-saturation techniques further improve suppression of the background signal, allowing for improved depiction of the veins, particularly when maximum intensity projection (MIP) algorithms are used.

The FSE MR venographic method can produce high-quality MR venograms in small extremity veins (e.g., calf and forearm) and is especially sensitive for depicting small branch veins (23).

In clinical practice, FSE sequences may have some value in cases where intravenous contrast is not recommended and other nonenhanced techniques provide inconclusive results. Improved visualization of small vessels in the calf by using FSE sequences has been described compared with time-of-flight (TOF) imaging (23).

Short tau inversion recovery (STIR) sequences can be effective in demonstrating slowly flowing blood as bright signal and can facilitate the depiction of vascular malformations (26,27).

In single-shot FSE imaging, all lines of k-space for a given slice are filled by multiple 180-degree refocusing pulses applied after a single excitation pulse (22). The “effective TE” in single-shot imaging represents the time between the excitation pulse and the particular TE of the echo train in which the central k-space line is acquired. Low-frequency lines in the center of k-space are typically sampled earlier in the echo train to provide adequate T2-weighted contrast and optimal signal-to-noise ratio (SNR). However, there is inherent blurring of single-shot images due to signal decay during the acquisition of the long echo train.

Half-Fourier techniques are commonly applied in conjunction with single-shot FSE sequences to further decrease acquisition time. These techniques rely on the inherent symmetry of k-space. By acquiring approximately 60% of the data to fill the k-space in half-Fourier techniques, such as half-Fourier single-shot turbo spin echo (HASTE, Siemens, Erlanger, Germany) or single-shot fast spin echo (SSFSE, General Electric Healthcare, Waukesha, WI), the remainder of the k-space can be mathematically calculated based on its symmetry. Since single-shot echo train imaging can be performed either with or without half-Fourier reconstruction, the acronym HASTE, which clearly connotes the use of half-Fourier reconstruction, will be used for the remainder of this chapter.

HASTE allows for very fast T2-weighted imaging, with most slices taking less than 1,000 milliseconds to acquire.
The quality of single-shot imaging is related to gradient performance. As a specific absorption rate (SAR)–intensive sequence, the number of slices obtainable in a multislice sequence can be constrained and compensated for with reduced flip angle (FA)–refocusing pulses.

Multislice acquisitions are acquired in a sequential manner, with each slice completely acquired before it proceeds to the acquisition of the next slice. This sequence provides a motion-insensitive strategy for imaging patients with limited or no breath-hold capacity. If the patient can cooperate, however, breath holding is still recommended to preserve sequential display of the anatomy.

HASTE images, like conventional T1- and T2-weighted SE images, characteristically demonstrate a flow void within the vessels as the excited protons exit the imaging slice. Slow flow may allow the blood protons to spend enough time within the imaging slice to provide intravascular signal. Since HASTE acquisitions are T2 weighted and the T2 of blood is relatively long, vessels with slow flow may demonstrate very bright signal in these images. For more uniform suppression of blood signal, so-called double inversion strategies can be effective (28,29).

Bright signal from slow flow should not be confused with thrombus, the latter typically demonstrating intermediate gray signal intensity on HASTE sequences (Fig. 24-11). While this may be a source confusion, in the authors’ experience, the negative predictive value of HASTE imaging when intravascular flow voids are present is excellent.

The inherent T2 contrast and short acquisition times make this sequence an excellent approach for an initial survey of the body as well as dynamic acquisitions during Valsalva maneuvers. Dynamic HASTE imaging is a widely available technique that provides a rapid screening for venous thrombosis without the need for intravenous administration of contrast.

Dynamic single-shot FSE images are acquired perpendicular to the vessel of interest by using the following parameters: TR 800 milliseconds; TE 62; FA 120; 192 ÷ 256 matrix; slice thickness = 6 mm. Images are obtained during end expiration at a relaxed status and following a sequence command. The patient is instructed to perform a series of seven gradually increasing and then decreasing Valsalva maneuvers (relax, mild push, moderate push, extreme push, moderate push, mild push, relax). These seven images can then be displayed as a single acquisition in a movie loop that facilitates visualization of changes in signal intensity and vessel size.

The Valsalva maneuver increases the intrathoracic pressure with a subsequent decrease in the venous return to the heart. Valsalva also decreases blood velocity and causes enlargement of peripheral veins. With gradually increasing Valsalva effort, normal veins become gradually larger, reflecting vessel distensibility, and increased signal intensity of the blood can appear as a result of slow flow and temporary pooling (Fig. 24-12). The demonstration of both increased signal intensity within the vessel lumen and expansion of vessel caliber in response to a Valsalva maneuver has excellent negative predictive value to exclude venous thrombosis (unpublished data). Thrombosed veins demonstrate intermediate gray signal intensity before and after Valsalva maneuvers on these images without apparent signal intensity change or vessel enlargement (Fig. 24-13).

Steady-state free precession (SSFP) imaging is a coherent steady-state technique in which a fully balanced gradient waveform is used to recycle transverse magnetization to preserve signal intensity in species with a long T2 (30). Examples of these techniques among different manufacturers include true fast imaging with steady-state precession (true FISP, Siemens, Erlanger, Germany), fast imaging employing steady state acquisition (FIESTA, GE Healthcare, Waukesha, WI), or balanced fast field echo (balanced FFE, Philips, Bothell, WA). The transverse magnetization is maintained in the steady state from one TR to the next by rewinding the gradient waveforms on all axes (30). The gradients are perfectly balanced, and the total gradient area is zero at the end of each cycle (30).

Image contrast depends on the T2/T1 ratio and is nearly independent of blood flow (31). The high T2/T1 ratio of blood yields high signal intensity on true FISP images without intravenous contrast material (30). The
rapid nature of this sequence provides diminished sensitivity to motion.

SSFP images obtained perpendicular to the vessel or vessels of interest are usually the most useful. Multislice acquisitions are acquired in a sequential manner, with each image acquired in approximately 1 second or less. This imaging approach is insensitive to respiratory artifacts and allows for a rapid evaluation of the entire chest, abdomen, or pelvis in a single breath hold. Short TRs and very short TEs are recommended to decrease artifacts related to T2* decay and off-resonance effects.

SSFP imaging has been proposed as an alternative to gadolinium-enhanced magnetic resonance angiography (MRA) for rapid assessment of patients with suspected
acute aortic dissection (32) and comprehensive evaluation of the entire hepatic vasculature in potential liver donors (33). It allows for thrombus to be visualized as intravascular filling defects with low signal intensity compared with the surrounding hyperintense blood (Fig. 24-14) and has been recently proposed as a noninvasive, fast approach for detection of venous thrombus in the veins of the abdomen, pelvis, and lower extremities (34).

Pulsatility artifacts in SSFP images (30) may cause heterogeneous signal intensity that can be misinterpreted as thrombus (35). These artifacts are more common at the level of the confluence of veins or in areas where the veins change abruptly in direction. Furthermore, venous thrombus can be missed on these images in patients scanned in the subacute stage (1 to 3 weeks after onset of symptoms) (35) (Fig. 24-15). This is presumably related to the thrombus having T2/T1 characteristics similar to that of the blood pool. Increased signal intensity on T1- and T2-weighted images of carotid thrombi induced in swine has been described with the relative increase in SI significantly
higher on T2-weighted images than that on the T1-weighted images occurring during the first 3 weeks (36). This phenomenon may account for increased signal intensity within the thrombus and false-negative results of SSFP imaging in patients who are imaged in the subacute stage (35). Nevertheless, SSFP imaging can be valuable in patients without intravenous access and suspected venous thrombosis in whom gadolinium-enhanced techniques are not feasible. This is particularly true in those patients with limited breath-hold capability due to the relatively motion-insensitive nature of this sequence compared with other unenhanced techniques (e.g., TOF, phase contrast [PC]). In a recent study, SSFP had a sensitivity of 66% and a specificity of 77% for the diagnosis of venous thrombosis when compared with gadolinium-enhanced T1 gradient echo images (35).

A potential advantage of SSFP imaging is the possibility to obtain cine acquisitions with the use of cardiac triggering (30). Cardiac triggering may reduce pulsatile artifacts and thus improve the homogeneity of the signal within vessels. To the authors’ knowledge, the impact of cardiac triggering in the diagnostic accuracy of MRV with SSFP sequences has not yet been explored. However, in their experience, this approach is particularly valuable in the evaluation of the SVC and segments of the IVC close to the right atrium. Cardiac triggering virtually eliminates pulsation artifact in these vessels and allows for continued display of intraluminal thrombus during the cardiac cycle (37) (Fig. 24-16). Extreme degrees of motion of venous thrombus within the SVC and IVC may correlate with higher risk for thromboembolic events.

A major accomplishment for MR imaging was the realization that TOF properties could be exploited for MR vascular imaging (38,39,40,41). TRs shorter than the longitudinal relaxation time (T1) of the stationary spins within the imaging slice lead to decreased signal from partial saturation effects (41). When unsaturated blood spins (not excited by the spatially selected radio frequency pulses) move from outside the slice into the imaging slice, they produce a much stronger signal than the partially saturated signal of the stationary spin within the imaging slice (41). This produces the “entry slice phenomenon” or “inflow enhancement,” in which signal intensity within the vessels is strong while signal in the stationary tissue is weak. For details on TOF imaging techniques, please refer to Chapter 2.

TOF optimization to depict slow flow within veins includes prescription of the imaging slices perpendicular to the vessel of interest and the use of thin sections (3 to 5 mm). A determination of vessel patency over a large FOV with TOF can be time-consuming. Fewer slices with interslice spacing reduce the total acquisition time. A demonstration of bright intraluminal signal in those slices will establish vessel patency. However, the relatively limited sampling could result in undetected, small partial thrombosis and in an inability to render a meaningful MIP reconstruction. This sparsely sampled TOF approach can be effective before gadolinium-enhanced MRV is performed for assessment of vessel patency and the direction of flow within the vein of interest.

Increased signal intensity caused by the entry-slice phenomenon is not directional. Systemic arteries and veins, when adjacent to each other, tend to run in opposite directions, but both appear bright on the slice imaged with TOF. A saturation band is used to selectively eliminate the signal from the flowing spins entering the slice from a chosen direction (42). For uniform suppression, the saturation
band is applied in a consistent, close proximity to the slice, referred to as a traveling saturation band.

MRV is performed by saturating the signal from the arterial spins. Placement of the saturation band, either above or below the imaging slice, depends on the anatomic location of the slice as it relates to the heart. Below the heart, the aortic signal is suppressed by placing the saturation band above the slice. In so doing, bright signal is selectively displayed within those vessels that have blood entering from below the slice (i.e., IVC, portal vein) (Fig. 24-17). This approach can be used for determination of both the vessel patency and its flow direction.

Above the heart, axial venographic TOF images are acquired with a saturation band applied below the imaging slice in order to eliminate the signal from arterial spins (Fig. 24-18). For the subclavian veins, sagittal images are needed for a perpendicular orientation with respect to the vessel orientation (Fig. 24-19). In order to suppress the arterial signal, the saturation band located medial to the imaging slice (left to the slice for the right hemithorax and right to the slice for the left hemithorax) will provide MRV TOF images of the subclavian veins.

Typical MR acquisition parameters for TOF imaging are TR = 25 to 30 milliseconds; TE = 6 to 10 milliseconds; FA = 30 to 45°; thickness 3 to 5 mm; matrix 128 ÷ 256, and application of flow compensation gradients. Low FAs (<20°) can lead to decreased suppression of the background signal as well as a poor SNR. Very high FAs (>45°) can cause saturation of the slower venous blood signal. A MIP algorithm applied to TOF images can generate venogram-like images, displaying the venous anatomy (and the extent of occlusion)
in a familiar and readily accepted manner (Fig. 24-20).

With TOF, venous thrombosis will appear as a hypointense filling defect within the hyperintense column of flowing blood (43,44,45,46). MR using 2D TOF has demonstrated good correlation with conventional venography in the diagnosis of venous thrombosis (44,47,48,49). Reported sensitivity and specificity are also high compared with duplex sonography (49). Pitfalls can result from in-plane saturation of the intraluminal signal simulating venous thrombus, particularly when blood is not flowing orthogonal to the imaging slice. Turbulent flow can manifest as an area of low signal intensity, seen often near venous confluences, also mimicking a filling defect. Subacute clot can demonstrate high signal intensity similar to that of the flowing blood, blending imperceptive with the normal blood signal (50). Thrombus that is not completely occlusive, especially small areas of clot adhering to the vessel wall, can be difficult to visualize by TOF (45).

PC is another technique that exploits the moving nature of blood to create an image. Spins moving in the presence of a magnetic field gradient accumulate additional phase shift (51,52,53). The amount of shift is directly proportional to the velocity of the spins and the strength of the gradient (52).

Flow-induced phase shifts arise only from the component of motion along the gradient direction. Thus, three orthogonal gradient directions must be employed in evaluating multidirectional flow (53). The strength of the gradient must also be carefully matched to the velocity of the blood being imaged.

Thus, it is desirable to know the velocity of the blood to be imaged before performing the imaging. Ideally, velocity-encoding values should be chosen to exceed the maximum expected velocity by about 25% (25). For PC imaging of the venous system, a velocity-encoding value of approximately 20 cm/second is typically used (25). Phase contrast venography (PCV) is rarely used in the body.

Conventional gadolinium contrast agents, distributed in the extravascular, extracellular fluid (ECF), offer excellent signal to noise during a delayed, equilibrium phase and yield intense homogeneous venous enhancement even after administration of a single dose of gadolinium. The relative safety and efficacy of gadolinium venography makes it an excellent choice for patients with certain degrees of renal insufficiency or an allergy to iodinated contrast media (55,56,57). Gadolinium venography has been useful to identify venous access for patients with renal failure (58). Recent concerns associating severe renal insufficiency, gadolinium contrast administration, and nephrogenic systemic fibrosis will need to be considered when balancing the risk:benefit ratio for gadolinium venography in this patient population (59,60,61).

Blood pool agents (e.g., MS-325, NC100150) offer prolonged and strong venous enhancement (62,63)
(Fig. 24-21). This capacity can significantly improve diagnostic conspicuity and contrast in iliocaval venous opacification compared with extracellular contrast media–enhanced MRV (63). The prolonged half-life may offer additional convenience with the possibility of contrast material administration preceding transfer of the patient to the imaging suite.

Three-dimensional T1-weighted gradient echo sequences cover a large area of anatomy in a single vascular phase. Twenty- to twenty-five-second acquisitions provide sufficient anatomic coverage during end expiration breath holds, when needed. Subsequent vascular contrast phases can then be obtained, while allowing the patient time to breathe in the interim. With reproducible breath-holding, subtraction technique can be used to maximize target to background contrast and to generate a venogram (64).

After a timing run and calculation of an appropriate delay time, the authors perform indirect MRV by using a 3D fat-saturated T1-weighted gradient echo sequence (TR = 4.2, TE = 1.7, FA = 25, 160 ÷ 256 matrix, slice thickness 1.5 to 2.5 interpolated, slab thickness 80 to 160 mm). Fat saturation should be avoided in the chest and neck, where inadvertent water suppression could eliminate the gadolinium signal (65). Contrast administration consists of a double dose (0.2 mmol/kg body weight) of gadopentetate dimeglumine (Magnevist, Berlex Laboratories, Wayne, NJ) through intravenous access in an antecubital vein. When upper extremity thrombosis is in question, the contralateral arm is injected.

With conventional extracellular-gadolinium agents, a biphasic administration yields good-quality enhancement of the blood pool. The total dose of gadolinium is administered as follows: 0.1 mmol/kg body weight at 2 cc/second immediately followed by 0.1 mmol/kg body weight at 0.8 cc/second, in a single injection. An additional 20 cc of saline flush is injected at 0.8 cc/second. Evaluation of the visceral veins, including the portal and renal veins, is typically performed with a single dose of gadolinium (0.1 mmol/kg body weight at 2 cc/second) with excellent results.

Bolus injection timing is essential when a selective venous study is desired. First, a “vein-free” arterial phase is captured by using a timing examination with administration of 1 cc of gadopentetate dimeglumine at a rate of 2 cc/second through a peripheral intravenous line in an antecubital vein, as previously described by Earls and associates (66). This approach ensures acquisition of the central region of the k-space during the middle of the administered bolus, thus guaranteeing appropriate arterial enhancement throughout that entire 3D data set. The delay time (Td), or the time between injection initiation and the scan start time, can be effectively calculated with the following equation:

Td = Tc +Ti/2 – Ts/2

where Tc is the circulation time, Ti is the injection time for the first component of the injection (administered at 2 cc/ second), and Ts is the scan acquisition time. Tc is the time between injection initiation and the peak arterial enhancement in the anatomic region of interest (ROI), and it is estimated by the timing run.

For example, if the contrast arrives to the abdominal aorta 20 seconds after initiation of the injection of 1 cc of gadolinium at 2 cc/second, then Tc equals 20. If the patient’s weight is 80 kg, a double dose (0.2 mmol/kg body weight) will be 80 ÷ 0.4 = 32 cc. A single dose (16 cc) is given intravenously at 2cc/second followed by a second dose (16 cc) at 0.8 cc/second in a single injection. The injection time for the first component of the injection given at 2 cc/second is 16/2 = 8 seconds. For an acquisition time of 24 seconds, the time delay for starting the acquisition would be:

Td = 20 + 8/2 – 24/2 = 20 + 4 – 12 = 12 seconds

In this example, acquisition of the first 3D data set (arterial phase) should be timed so that the acquisition is initiated 12 seconds into the gadolinium administration. Breathing instructions are typically given before starting the acquisition. Hyperventilation prior to the breath-hold procedure improves the patient’s performance. In the authors’ experience, two cycles of the command “breath in, breath out” take approximately 6 seconds and yield good results. Therefore, in the theoretical patient, two sets of breathing instructions would be started 6 seconds after the initiation of the contrast injection, and the acquisition would be started immediately after completing the breathing instructions (12 seconds after initiation of the injection of contrast).

After the “venous-free” acquisition is obtained during the arterial phase, two additional acquisitions are obtained at 40 seconds and 90 seconds delay after the arterial peak. Typically, delayed blood pool acquisitions are used for making the diagnosis.

The arterial (A) acquisition can be used as a mask that is subtracted from the delayed blood pool phase, the latter containing both arteries and veins (AV) (64). MIP reconstructions of the subtraction result, AV – A, resemble conventional venography images (64) (Fig. 24-22).

Three-dimensional T1-weighted gradient echo sequences provide accurate reconstructions in any plane, facilitating the evaluation of tortuous veins and veins that are oriented parallel to the acquisition plane.

Direct gadolinium-enhanced MRV (67) uses a very low concentration of gadolinium chelate to avoid first-pass susceptibility effects that can obscure key information. Diluted gadolinium contrast media (3 cc in 60 cc of saline for a 1:20 solution) is injected directly through peripheral intravenous access in the affected extremity. Bilateral intravenous access may be needed in both upper extremities for the assessment of the SVC in order to minimize mixing artifacts from nonopacified blood entering the SVC from the contralateral extremity. A tourniquet in the lower extremities helps to prevent enhancement of the superficial venous system (67).

A 3D fat-saturated T1-weighted gradient echo sequence as described earlier is used when direct MRV is performed during the administration of the contrast media. The acquisition can be started 7 to 10 seconds after the initiation of contrast media delivery. Multiple acquisitions may be obtained over time to improve visualization of small veins and veins with slow flow enhancing in a delayed fashion.

This technique improves visualization of patent veins compared with indirect MR techniques. In addition, lack of arterial enhancement facilitates the interpretation of these examinations as well as multiplanar and volumetric reconstructions (i.e., MIP, volume rendering [VR]). The diagnosis of venous thrombosis is made based on the nonvisualization of a particular venous segment. However, similar to conventional venography, early acquisitions during the administration of contrast may fail to demonstrate venous pathology or can be difficult to interpret. Delayed imaging after the direct MR examination is completed may yield images similar to those of the indirect MR technique. These are useful for the visualization of venous pathology, including occluded veins (Fig. 24-23).